Published in Solid State Technologies, May 1998, Page 79

Introduction:

Written by: Richard Bothell and Justin Bothell, Atlas Technologies; Glen Tisdale, Judith Offerle, UHV Aluminum Company

300mm wafers with

Achieving sub-micron feature sizes on a production basis requires an understanding of the complete processing environment including the vacuum reactor itself. Molecular level chemical contamination will demand that processes, PVD, CVD, Etch, Ashing, lithography etc., all consider the impact of the vacuum environment on the process. The vacuum system base pressures is a critical parameter in determining the environmental contribution to wafer contamination. Actual processing levels may be decades above the base pressure, but the base pressure establishes the environmental noise of the system.

Ultra high vacuum (UHV) base pressures reduce the contamination contribution of the vacuum environment. For example at high vacuum levels (~10 -9 Torr) mono-layer formation takes hours (any vacuum text). In order to control molecular level contamination of sub-micron features, UHV base pressure environments must be considered.

The current concern with producing UHV environments with present aluminum vacuum systems technology is the cost of the massive pumping required to overcome o-ring seal leaks and vacuum surface out gassing. And to consider stainless steel UHV systems means that cycle times will be increased due to bake out and pump down times. Stainless is a notorious water sponge at these levels and iron and chromium contamination is an issue as well. In addition stainless systems are heavy and require a large foot print of valuable clean room real-estate.

Aluminum UHV Systems

Aluminum vacuum systems have been used in semiconductor production since the beginning. Its’ low cost of fabrication and over all cost of ownership are well understood. Aluminum has become the dominate vacuum system material for many low and high vacuum cluster tools and other processing reactors.

Until recently, the technology required to process aluminum into a UHV system has not been well understood. However, developments over the past 15 years in physics laboratories world wide now make this possible. The methods, process, and components used to build large UHV systems for particle accelerators and synchrotrons are now being adapted for semiconductor processing applications . There are three primary technical challenges that have been met in this endeavor. They are the development of aluminum surface treatment schemes and outgassing reduction, the development of ultrahigh vacuum capable aluminum welds, and finally the development of robust metal sealed aluminum flanges.

Three Technical Advances that Allow Ultra High Vacuum Performance with Aluminum Systems

To benefit from the many superior physical properties of aluminum for UHV systems, relatively minor changes in existing manufacturing processes and components are required. Current designs of aluminum process tools can be up graded to improve vacuum performance. In particular this means improving the outgassing and permeation performance of these systems. [1]

Aluminum Surface Treatment and Outgassing Reduction

Like untreated or machined stainless steel the off the shelf mill grade surface of aluminum must be striped and prepared for UHV application. Extruded and rolled plate stock can not be used directly it tends to be covered with a porous oxide that it relatively thick, typically 150 Å -200 Å . One of the primary reasons for this is that when aluminum is extruded, alloy impurities segregate to the surface where they combine to produce oxide impurities. Further, when aluminum is machined thick oxides can quickly form on the raw machined surface. These surfaces must be properly treated to obtain the thin dense uniform oxide needed to attain UHV. The left half of Figure 1 illustrates the morphology of a typical untreated aluminum surface.

Aluminum Surface Morphologies

For making aluminum UHV compatible, what is required is some means for ridding the surface of any contaminating constituents and then producing a surface oxide that is thin and non-porous.

Ishimaru, of KEK Japan, was the first to apply a concentrated effort toward solving these problems in his efforts to build aluminum vacuum systems for particle accelerators. He employed two primary processing techniques, one for extrusions and the other for machined components. For extruded surfaces Ishimaru would extrude aluminum into an argon-oxygen atmosphere. For machined components he used ethanol as a machining fluid. [2] Both of these techniques enable the aluminum to form a thin native oxide which is dense and non-porous as illustrated in the right half of Figure 1. The results for these techniques have been highly successful with achieved outgassing rates of

These techniques have been refined during the past 15 years to simplify their implementation. Through work at a number of particle acclerator facilities it has become apparent that surface cleaning with simple alkaline solutions is as effective as argon oxygen extrusion and ethanol machining. [4] These alkaline solution cleaning will remove impurity oxides as well as the top layer of aluminum oxide. Any contaminants introduced from machining and that have been accumulated in the aluminum oxide layer will be removed when the aluminum oxide layer is removed. Additionally when the top layer of the aluminum oxide is removed by these chemicals a thin, dense and non-porous native oxide will form on the newly exposed aluminum surface.

Aluminum Vacuum Flanging

Aluminum high vacuum systems have typically been sealed with elastomeric o-rings. Elastomers both outgas and are permeable to atmospheric gases. Permeation occurs because atmospheric gases are able to dissolve into the elastomer and then diffuse through the material where they are released into the vacuum chamber. Unlike outgassing, permeation cannot be eliminated by baking the system. Eliminating o-rings from the system can significantly reduce pumping requirements, and cost.

Furthermore research has been conducted to understand the influence of common vacuum impurities on aluminum sputtered films. One study observed the impact of nitrogen contamination generating from atmospheric permeation through O-ring seals 1 . The upshot of this particular study is that nitrogen will readily incorporate itself into an aluminum film while the film is being grown and this nitrogen impurity will noticeably impact the electromigration resistance.

There are two methods for eliminating o-ring gas permeation, both of which will be required in future aluminum ultrahigh vacuum systems. The first is an old solution and involves using differentially pumped seals.

Differentially Pumped Sealing Geometry

In a differentially pumped sealing geometry the main vacuum chamber is isolated from the atmosphere by a secondary evacuated channel which is separately pumped. Since permeation is proportional to the pressure differential, even a rough vacuum in the pump channel can reduce the permeation by several orders of magnitude. The drawback to this sort of scheme is that it requires an auxiliary channel for each vacuum flange and a pump to evacuate these channels. However for ports that require quick access such as sputter target ports differential pump schemes do provide a solution that can provide the required vacuum performance.

The other solution is to replace the elastomeric seals with metal sealed flanges. The early metal sealed flanges were composed entirely of aluminum and had a ConFlat â style knife edge which could seal an aluminum gasket. Typically the knife edges of these flanges are coated with titanium nitride to strengthen the knife edge. These flanges were subject to three important limitations. First the knife edge, since it is composed of aluminum, is readily damaged because of the softness of the underlying aluminum. Second, overheating the flanges during weld up or while heat cycling, anneals the aluminum, making the knife edge more subject to failure. The third issue is more subtle. Because most vacuum components are stainless steel a typical aluminum flange will serve as a mount for a stainless component. These types of joints do not tolerate temperature cycling well because the differential thermal expansion between the aluminum and stainless steel will loosen the knife edge seal to the gasket and cause the seal to leak. The standard solution for this type of problem has been to provide a temperature differential during system bakeout between the two materials so that the thermal expansions are matched.

A creative solution to this problem has been developed by Atlas Technologies [5] and involves the construction of a bi-metal flange. This flange consists of a stainless steel surface which contains a ConFlat â knife edge bonded to an aluminum nipple used for weld up and is illustrated in Figure 3.

Bimetal Flange Detail

It provides a robust sealing surface which is composed of stainless steel, yet it may be readily welded to an aluminum vacuum system. Because the stainless and aluminum surfaces meet at the bond rather than at the gasket, the sealing surface are not subjected to differential thermal expansion and consequent leaking. Flanges such as this will serve as ready replacements for elastomeric seals in flanges that do not need to be routinely opened.

Aluminum Welding

The aluminum surface of a vacuum system can be treated so that the outgassing is minimized and the system sealed with flanges to minimize permeation. However the flanges must still be joined to the system in a fashion that allows for the system to achieve superior vacuum performance. In the case of aluminum vacuum hardware this implies that the flanges must be welded to the remainder of the system with welds that are non-porous so as not to create virtual leaks and mechanically strong to maintain the structural integrity of the system.

Producing non-porous welds in aluminum is difficult for a number of reasons. [6] First at high temperatures hydrogen becomes highly soluble into molten aluminum. When the aluminum subsequently cools, the hydrogen bubbles through the melt and forms pores in the material. Second, aluminum oxide has a high melting temperature which implies that any aluminum oxide that is present at the weld surface will collect within the weld and form a defect layer. Third, because aluminum has a high thermal conductivity the heat of the weld will tend to be dissipated within the material bulk. And finally, because aluminum shrinks when it solidifies aluminum welds will be more subject to cracking than their counterparts in steel.

Presence of the effects above does not imply that aluminum welds cannot be produced, it does however mean that producing them will require adequate material preparation and tight control over the processing parameters to insure repeatability. To meet the challenges of welding aluminum the vacuum community has developed automated TIG welding processes, which when coupled with careful material preparation yield reliable ultra high vacuum welds. Figure 4 presents an example of automated welding equipment.

Aluminum as a Vacuum Chamber Construction Material

When learning how to use aluminum for UHV, a vacuum system engineer experienced in stainless steel UHV system designs will be first skeptical, then interested, and then excited by the powerful capabilities that the properties of aluminum offer. Here we shall discuss those properties of aluminum that are of interest to vacuum designers, in our description of these properties we use stainless steel as a convenient benchmark.

Alloys [7]

Aluminum comes in a wide range of alloys which vary as to characteristics. Wrought aluminum alloys are suitable for vacuum chamber fabrication and are designated by a four digit number which is followed by a temper specification. The first digit specifies the primary alloy composite as indicated in Table 1.

Table 1 : Aluminum Alloy Designations

All of the wrought alloys have been used for vacuum chamber fabrication except for the 7000 series which has zinc which has a high vapor pressure at low temperature. The 2000 series alloys are highly weldable. [8] The 6000 series, particularly 6061, and 6063 has been used for ultrahigh vacuum systems. [9]

Machinability

One of the overriding impetuses for fabricating vacuum systems from aluminum is that it is inherently more machinable than materials, such as stainless steel, which have been traditionally used for vacuum component fabrication. The machining cost for 300 series stainless steel is 5.5 times that of aluminum [10].

As indicated previously this ease of machinability is the reason that aluminum has become the material of choice for fabricating semiconductor equipment. Cluster tool components are machined from a single plate of aluminum. These tools are exceptionally rigid, and have a minimal vacuum surface area and occupy a minimal floor space considering their throughput.

Mechanical Properties

Typical elastic modulii for aluminum alloy 6061 T6 and 304 alloy stainless steel are respectively and . If these values are used in standard mechanical formulae for standard geometry’s the ratios of critical thickness for the two materials are: [11]

Here , and are respectively the minimum thickness ratios to avoid buckling in flat plates, long cylinders and short cylinders.

What is noteworthy here is that the ratios are close to unity. This implies that an aluminum vacuum system will not require parts that have appreciably greater thickness than similar ones manufactured from stainless steel. This challenges the conventional wisdom that aluminum systems will have significantly more bulk than those constructed from stainless.

Thermal Conductivity

Aluminum has a thermal conductivity that ranges between 170 watt / m K and 230 watt / m K depending on the alloy. Stainless steels by contrast have thermal conductivity’s that are between 14 watt / m K and 16 watt / m K.

High thermal conductivity is an advantage when designing systems that require temperature cycling. For vacuum systems, this is the case for systems that must be baked to achieve ultrahigh vacuum. An aluminum chamber may be baked and then cooled much more rapidly in development of ultrahigh vacuum than a stainless chamber. Furthermore its’ conductivity allows for a complete bake out with out the problems of re-condensation of gasses on cool spots common in stainless systems.

Weight

Aluminum is 1/3 the weight of stainless steel. The cost burden associated with weight starts with raw materials handling and progresses through out the manufacturing process including shipping, installation and even architectural engineering and construction.

Magnetic Properties

Aluminum is not magnetic whereas stainless steel, being essentially an alloy of iron exhibits residual magnetism. For applications that involve charged particle beams the absence of magnetic properties in aluminum is of advantage because it implies that the vacuum system will not modify the fields from the beam control magnets.

Radioactivity

Aluminum in comparison to stainless steel has a much more rapid decay of induced radioactivity. Generally if both types of materials are bombarded with the same flux of charged particles the residual radioactivity will be between one and two orders of magnitude less for an aluminum sample than for an identical sample of stainless steel. [12]

The nuclear half life of elements that make up stainless suggests that a -particle contamination is always present in stainless steel and a possible source of circuit damage.

Corrosion Properties

The corrosion of both aluminum and stainless steel alloys in reactive gases is a complicated process. However much experimental work has been performed on various alloys in different reactive gaseous environments. The general thrust of these results is that both aluminum and stainless steel are subject to attack by reactive gases, that halogen containing species are typically the most damaging and that the corrosion activity of any given compound is usually no worse than that of its halogen component alone. [13]

Importantly aluminum is not a worse corroder but simply one with different reaction dynamics.

Outgassing Properties

Vacuum surface gassing is a primary source of process contamination. Water vapor is the most significant contaminate placing heavy demands vacuum pumping. Further more metal contamination is a major yield-limiting factor for silicon IC production and iron is one of the most significant contaminates. Outgassing rates of

This level of outgassing performance has not always been achievable with aluminum. Only recently have techniques been developed allow these levels of performance to be achieved repeatably. This improvement in outgassing performance has been one of the principal breakthroughs that has allowed aluminum to become a competent material for the construction of ultrahigh vacuum systems.

Conclusions

The next generation aluminum vacuum systems will be capable of Ultra High Vacuum for wafer processing and for base pressure system conditioning. Three technical advances have made this possible and allow the many physical properties of aluminum to be advantaged. Surface treatment pacification, welding process equipment, design know how, and metal sealed flange technologies have made this possible. The impact of ultrahigh vacuum for contamination free manufacturing processing has yet to be determined in a quantitative fashion, however the current studies indicate that it will be an essential component of semiconductor materials processing and control. Aluminum ultrahigh vacuum systems are currently being integrated into leading edge semiconductor processing equipment. This will happen with the deployment of processing systems.

Acknowledgments

We would like to pay our respects to Dr. Hajime Ishimaru deceased. Dr. Ishimaru pioneered the use of aluminum for UHV , we are all forever in his debt The authors would also like to acknowledge James Garner of SMC Corporation for many helpful discussions. Much of what is written here bears his imprint. Additionally we would like to thank the vacuum group of the Advanced Photon Source at Argonne National Laboratory which includes John Noonan, Joe Gagliano, George Goeppner, Richard Rosenberg and Dean Walters. They are responsible for much of the foundation work in this area..

Biographical Information

Richard Bothell
Richard Bothell is President of Atlas Technologies. Atlas Technologies supplies aluminum/stainless and copper/stainless bonded flanges, Bothell is a co-inventor of the Atlas Flange.

Jed Bothell
Jed Bothell is Vice President of Atlas Technologies. Atlas Technologies supplies aluminum/stainless and copper/stainless bonded flanges, Bothell is a co-inventor of the Atlas Flange.

Iron contamination

Further work

This study when taken as a whole indicates that vacuum contaminants will play an important role in semiconductor device performance. The authors conclude by suggesting the need for an enhancement in the system vacuum performance. We should note that this study is very specific: it applies to sputtering and aluminum films in particular

Technically this involved two separate experiments. To determine the adsorption kinetics, a freshly grown aluminum film was exposed to several isotopes of common vacuum contaminants. Isotopes were employed to facilitate detection. The completed films were then analyzed using several standard surface analytical techniques, primarily secondary ion mass spectroscopy. The net result was a set of curves that described the accumulated impurity concentration as a function of impurity exposure. As a second phase a number of sputtered films were produced in the presence of an isotope background. These were then examined again as to impurity content. In addition the films produced were benchmarked for certain standard performance indicators including electromigration resistance and sheet resistance.

Attention must be given to the vacuum system design, making quantitative as to the impact predictions is that neither of these questions have been dealt with in a quantitative fashion for most of the processes that are employed in semiconductor processing. A notable exception is a study which deals with the sputtering process. [15] This study was designed to answer both the questions regarding incorporation of impurities into semiconductor devices and the phenomena which these impurities in turn generate.

Until these sorts of studies have been produced for other process technologies the requirement for enhanced vacuum performance will be based merely on general arguments like those suggested previously.

Base pressures will follow the technology road map to ultra high vacuum base pressures.

Process Impact of Aluminum Ultrahigh Vacuum Systems

From the previous discussion it is evident that aluminum vacuum systems may be constructed to achieve ultrahigh vacuum performance. The cost of ownership for these systems has not been investigated. The many physical properties of aluminum as discussed above, suggest significant advantages over stainless systems for processing application. One can speculate as to the impact of these systems but little quantitative data that illustrates directly the benefits of ultrahigh vacuum processing has been produced. Cost of the installed system, processing cycle time, throughput, and yield are all part of the cost of ownership equation. Each of these categories must be quantified with additional research to understand the impact and use of aluminum for ultra high vacuum processing.

To more fully understand the previous assertion we need to place ultrahigh vacuum processing in context. There are two primary sources of contamination in any semiconductor processing operation. These are the starting materials and the processing environment. The process materials include such things as sputtering targets and reacting gases. The processing environment includes the vacuum environment. Both the starting materials and the vacuum environment may be treated as impurity sources. Consequently in discussing vacuum systems we concerning ourselves with materials purity issues. In particular to measure the benefit of improved vacuum performance we need to understand how the improved processing purity enables the development of improved devices.

It is possible to formulate general back of the envelope arguments that address these questions. These are very straightforward and involve contrasting the amount of impurity introduced by one mode such as the input gas versus the amount of impurity introduced by another mode such as the vacuum. For example given that a gas has a contamination of one part per billion of oxygen and is used in a 1 Torr process, it implies that the background pressure of oxygen in the vacuum system will need to be to match the gas purity. These types of arguments give a rough estimate as to the level of vacuum performance that will be required to match the performance of the other impurity sources.

However the previous arguments do not give understanding as to the microscopics involved regarding impurity additions. These microscopics involve two issues. The first concerns the mechanisms by which impurities in the environment and the materials get incorporated into the devices being constructed in the system. For example in the presence of an extraneous gas, the relevant issues concern the quantities that collect at the device interfaces and how these will get distributed through the device as it is processed. The second issue concerns the phenomena that will be produced by various impurity atoms. For example when excess oxygen is present in a thin film, the questions to be addressed concern the changes that this impurity will have on the properties of the film such as the electromigration resistance and the sheet resistance.

The difficulty in making quantitative predictions is that neither of these questions have been dealt with in a quantitative fashion for most of the processes that are employed in semiconductor processing. A notable exception is a study which deals with the sputtering process. [16] This study was designed to answer both the questions regarding incorporation of impurities into semiconductor devices and the phenomena which these impurities in turn generate.

This research was designed to understand the influence of common vacuum impurities on aluminum sputtered films. Technically this involved two separate experiments. To determine the adsorption kinetics, a freshly grown aluminum film was exposed to several isotopes of common vacuum contaminants. Isotopes were employed to facilitate detection. The completed films were then analyzed using several standard surface analytical techniques, primarily secondary ion mass spectroscopy. The net result was a set of curves that described the accumulated impurity concentration as a function of impurity exposure. As a second phase a number of sputtered films were produced in the presence of an isotope background. These were then examined again as to impurity content. In addition the films produced were benchmarked for certain standard performance indicators including electromigration resistance and sheet resistance. The upshot of particular study is that nitrogen will readily incorporate itself into an aluminum film while the film is being grown and this nitrogen impurity will noticeably impact the electromigration resistance.

This study when taken as a whole indicates that vacuum contaminants will play an important role in semiconductor device performance. The authors conclude by suggesting the need for an enhancement in the system vacuum performance. We should note that this study is very specific: it applies to sputtering and aluminum films in particular.

Until these sorts of studies have been produced for other process technologies the requirement for enhanced vacuum performance will be based merely on general arguments like those suggested previously.

[1] O’Hanlon, John. Ultrahigh Vacuum in the Semiconductor Industry , Journal of Vacuum Science and Technology A 12 , p. 921.

[2] Ishimaru, Hajime. Ultimate Pressure of the order of in an Aluminum alloy Vacuum Chamber . Journal of Vacuum Science and Technology A 7. p. 2439.

[3] Chen, J.R. et al. Thermal Outgassing from Aluminum Alloy Vacuum Chambers . Journal of Vacuum Science and Technology A 3 . p. 2188.

[4] Rosenberg, Richard et al. X-Ray Photoelectron Spectroscopy Analysis of Aluminum and Copper Cleaning Procedures for the Advanced Photon Source . Journal of Vacuum Science and Technology A 12 . p. 1755.

[5] Atlas Technologies, Port Townsend, WA, www.atlasbimetal.com

[6] Goeppner, George. TIG Welding of Aluminum Alloys for the APS Storage Ring – A UHV Application . Argonne National Laboratory Light Source Note 254.

[7] Brady, George S. and Henry R. Clauser. Materials Handbook, Thirteenth Edition . Mc-Graw Hill, Inc. 1991.

[8] Garner, James. Aluminum Based Vacuum Systems. Handbook of Vacuum Technology. ed. Hoffman, Thomas and Singh. Academic Press, Harcourt Brace and Company, 1997.

[9] Goeppner, George. APS Storage Ring Vacuum Chamber Fabrication . Vacuum Design of Synchrotron Light Sources, American Vacuum Society Series 12. ed. Yeldez G. Amer. American Institute of Physics. 1990.

[10] Profile Milling Requirements for Hard Metals 1965-1970 . Report of the Ad Hoc Machine Tool Advisory Committee to the Department of the U.S. Air Force, May 1965.

[11] Garner. Academic Press. 1997.

[12] Ishimaru, Hajime. All aluminum alloy ultrahigh vacuum system for a large scale electron positron collider . Journal of Vacuum Science and Technology A2. p. 1170.

[13] Garner. Academic Press. 1997.

[14] Ishimaru. Journal of Vacuum Science and Technology A2. p. 1170.

[15] Hashim, Imran et al. Vacuum Requirements for next wafer size physical vapor deposition system . Journal of Vacuum Science and Technology A 15 . p. 1305.

[16] Hashim, Imran et al. Vacuum Requirements for next wafer size physical vapor deposition system . Journal of Vacuum Science and Technology A 15 . p. 1305.

Ultimate Pressure of the Order of 10 -13 Torr in an Aluminum Alloy Vacuum Chamber Hajime Ishimaru, Journal of Vacuum Science Technology A 7(3), 2439-2442 1989

Thermal Outgassing of Aluminum Alloy Vacuum Chambers R. Chen, K. Nanushima, H. Ishimaru; Journal of Vacuum Science Technology A 3(6), 2188-2191 (1985).

Correlation of Outgassing of Stainless Steel and Aluminum F. Dylla, D. M. Manos and P. H. LaMarche, Journal of Vacuum Science Technology A 11(5), 2623-2636 (1993).

Outgassing from Aluminum Surface Layer Induced by Synchrotron Radiation Ota, K Kanazawa, M. Kobayahi, & H. Ishimaru; American Vacuum Society (1996).